Among U.S. men, bladder cancer is the fourth most commonly diagnosed malignancy, excluding skin cancers. 1 Bladder cancer mortality rates tend to be higher in more affluent countries, but variations in rates across geographic regions within countries, including the United States, are often substantial. 1 Mortality rates for bladder cancer, compiled over five decades, are elevated for men and women in both urban and rural areas 2,3 of the northeastern United States. 4–6 At least some of the geographic patterns observed in bladder cancer incidence and mortality may be explained by patterns of cigarette smoking. 7,8 While male mortality rates of lung cancer in the Northeast have changed from being among the highest in the nation to below the U.S. average over the past two decades, 9 bladder cancer mortality rates in the Northeast have remained high over the same time periods. 6
Investigators have hypothesized that employment in certain industrial occupations in the Northeast may explain a portion of the elevated mortality rates for bladder cancer. 2,3 Industries using organic chemicals have, in the past, exposed individuals to bladder cancer carcinogens, such as aromatic amines. 10 While exposures to known bladder carcinogens have been curtailed in chemical industries, 10 the carcinogenic effects of these exposures may not be manifested until decades later. 11 Other occupational exposures, such as in textile or rubber industries, may also increase the risk of bladder cancer. 3,10
In addition to occupational and smoking exposures, some of the geographic variation observed within the United States may be related to other factors, such as nutrition. We examined potential non-occupational factors, particularly dietary and lifestyle factors, that may influence bladder cancer rates in the Health Professionals Follow-up Study. This population is ideal in focusing on non-occupational exposures because all participants are male health professionals, and any regional differences in rates cannot be attributed to job-related exposures. In addition, the impact of educational and income disparities are reduced, allowing us to better isolate specific effects of dietary and lifestyle factors.
The Health Professionals Follow-Up Study is a prospective cohort initiated in 1986, when 51,529 predominately white men aged 40–75 years answered a detailed questionnaire by mail on diet and medical history. This cohort consists of dentists (57.6%), veterinarians (19.6%), pharmacists (8.1%), optometrists (7.3%), osteopathic physicians (4.3%), and podiatrists (3.1%). All 50 states were represented and no exclusions were made by race. Every two years, follow-up questionnaires were mailed to surviving cohort members to update data on medical conditions and exposures. This investigation was approved by the Human Subjects Committee of the Harvard School of Public Health.
To form the cohort for analysis, we excluded 2074 men with cancers (excluding non-melanoma skin cancer) diagnosed before 1986. The remaining 49,455 men were eligible for follow-up. Follow-up rates for this cohort averaged 94% per follow-up cycle during the six biennial cycles between 1986 and 1998. The National Death Index was used to determine vital status for nonrespondents, and the remaining nonrespondents were assumed to be alive and at risk for bladder cancer incidence.
At the beginning of the study, each participant provided his zip code, allowing us to extract population density data (urban and rural) from the 1990 Census. “Urban” areas in the 1990 Census are defined as “comprising all territory, population, and housing units in urbanized areas and in places of 2,500 or more persons outside of urbanized areas”. Since zip codes can encompass large areas and include both urban and rural settings, we created three categories: zip codes entirely rural, zip codes exclusively urban, and zip codes where the population is both urban and rural. When examining associations between urbanization and bladder cancer within smaller regions, we combined the rural category with that of rural/urban because of small numbers in the rural only category.
To assess dietary intake, we used a validated 131-item semiquantitative food frequency questionnaire (FFQ). 12 The baseline dietary questionnaire was administered in 1986 and dietary intake information was updated in 1990 and 1994. The questionnaire assesses average frequency of intake over the previous year. For each man, we calculated caloric and nutrient intake by multiplying the reported frequency of each food item by the caloric or nutrient content for the specified portion size. Food composition data were primarily based on the U.S. Department of Agriculture. 13
A reproducibility and validity study of the FFQ among 127 men in this cohort indicated that most foods are reported reasonable well; Pearson correlations between the average intake assessed by two 1-week diet records completed 6 months apart and 1-FFQ (completed after the diet records) ranged from 0 .31 for pie to 0.92 for coffee (mean = 0.59). 14
Assessment of Nondietary Factors
At baseline, and biennially thereafter, men provided information on their state of residence, current smoking status, exercise habits, weight, height, and medication use. The baseline questionnaire provided detailed information on past smoking habits, time since quitting, average number of cigarettes smoked per day before age 15, and at ages 15–19, 20–29, 30–39, 40–49, 50–59, and 60 and older. Total pack-years of smoking was derived to incorporate all past smoking experience. One cigarette pack-year is equivalent to having smoked one pack or 20 cigarettes per day for an entire year.
Participants were asked to report specified medical conditions including cancers that occurred in the two-year period between each follow-up questionnaire. Whenever a participant (or next-of-kin for decedents) reported a diagnosis of bladder cancer, we asked for permission to obtain related medical records or pathology reports. When permission to obtain medical records was denied, we attempted to confirm the initial cancer report and date of diagnosis with an additional letter or phone call. If the primary (or ancillary) cause of death of a participant, as reported on their death certificate, was a previously unreported bladder cancer, we contacted family members to obtain permission to retrieve medical records, or to confirm the diagnosis of bladder cancer. Bladder cases previously unreported during follow-up and subsequently detected from death certificates constituted 7% of the total.
The endpoints in this study were 328 bladder cancer cases newly diagnosed between 1986 and January 31, 1998. We obtained medical records for 281 of the cases and were able to confirm the diagnosis of bladder cancer in all of these. When medical records were available, stage could be determined in 88% of the cases; remaining records did not have sufficient information. Self-reported cases for which we could not obtain medical records were kept in the analyses after verifying that similar results were obtained after excluding those cases. Based on pathology reports, over 90% of bladder cancer cases were transitional cell carcinomas and over 97% were localized in the bladder. Cases with stage II (tumor invades muscle) or greater were grouped as “advanced” cases. We included deaths from bladder cancer (if reported as the primary cause of death) in the “advanced” category, as these cancers would have been metastatic (N = 22). Stage 0a or I were classified as “nonadvanced” cancers. Carcinoma in situ tumors (N = 6) were not grouped in either two categories due to their aggressive characteristics; even though they are superficial lesions, they have a high risk of progression. 15
We computed person-time of follow-up for each participant from the return date of the 1986 questionnaire to the date of bladder cancer diagnosis, death from any cause, or January 31, 1998, whichever came first. In the main analysis, exposure categories were determined by the 1986 questionnaire (including residence location), except for age and current smoking status, which were updated every two years in all analyses. U.S. states were grouped into four geographic regions, according to the U.S. Census definitions: West (includes pacific), Midwest, South, and Northeast. The incidence rate for each region was calculated as the number of bladder cancer cases divided by the person-time of follow-up. The relative risk (RR) was computed as the incidence rate for one region divided by that of another region.
For multivariate analyses, we used pooled logistic regression 16 with two-year time increments to adjust for age (5-year categories), pack-years of smoking (7 categories including never smokers), and current smoking status (yes/no). This approach is asymptotically equivalent to the Cox regression model with time-dependent covariates, given short time intervals and low probability of the outcome within the interval. 16 Participants who died or were diagnosed with bladder cancer during a two-year cycle were censored at the end of that two-year period.
Regional variations of lifestyle characteristics and dietary intakes of the Health Professionals are summarized in Table 1. Very few cohort members were either African- or Asian-American; African-Americans were evenly dispersed over the different regions, but a larger fraction of the Asian-American participants lived in the West. There was substantial regional variation by religion. A substantial percent of participants located in the Northeast were Ashkenazi Jewish (36.9%), whereas in the West only a small percent of the participants were of this ethnicity (9.9%). Percent of current smokers and number of pack-years of cigarettes smoked were relatively lower in the West and the Northeast compared with other regions. Routine physical exams, routine endoscopy and prostate specific antigen (PSA) tests were slightly more common in the Northeast than in other regions. Dietary intakes of the three major energy-sources were similar across the four regions, and no substantial differences were observed for vitamins E and C intakes.
Compared with the West, incidence of bladder cancer was 71% higher in the Northeast (Table 2). When we further divided the Northeast region into New England and the Middle Atlantic, bladder cancer risk was 81% greater in the Middle Atlantic region, but the relative risk in New England was similar to the South, compared with the West (Table 2). Only two of the three states in the Middle Atlantic region (New York State and New Jersey) had elevated risks when each state was compared with all other (non-Middle Atlantic) states (age-adjusted RR NY = 1.48, 95% CI = 1.07–2.05; RR NJ = 1.57, 95% CI = 0.98–2.53; RR Pennsylvania = 1.09, 95% CI = 0.68–1.73).
Table 3 provides case numbers and actual incidence rates of bladder cancer for each region by age strata. Bladder cancer rates are consistently higher in the Northeast across age groups.
Controlling for pack-years of cigarette smoking and current smoking status attenuated the differences in risk of bladder cancer across regions, but the relative risk of the Northeast when compared with the West remained elevated (Table 2). We examined the possibility of confounding by additional smoking variables, including time since stopped, duration, pack-years, and age at start (to determine latency period). Although all these variables predicted risk of bladder cancer (Table 4), substituting these variables for pack-years in the models did not further attenuate the relative risks of the regions (data not shown). In a stratified analysis, ever smokers in the Northeast (compared with the West) had a relative risk of 1.85 (95% CI = 1.23–2.77), which did not vary by pack-years of smoking, and never smokers had a relative risk of 1.44 (95% CI = 0.73–2.86).
Race, Religion, and Profession
In this cohort, dentists were at a slightly higher risk of bladder cancer than all other health professions (RR = 1.20; 95% CI = 0.96–1.50, adjusting for age and smoking). Compared with other religions, Protestants and other Christians had a lower risk of bladder cancer. Southern European descendants had an elevated risk of bladder cancer compared with other Caucasians (RR = 1.44; 95% CI = 1.12–1.85, adjusting for age and smoking). Controlling for marital status, health profession, race/ancestry, or vasectomy did not substantially alter the relative risks for bladder cancer by region. Controlling for religion slightly attenuated the relative risk in the Northeast region (RR = 1.54, 95% CI = 1.09–2.16, when simultaneously controlling for smoking history). Given the substantial number of men for whom we did not have information on religion (as this was not asked until 1998), we performed an additional analysis among participants with data on religion (RR = 1.74, 95% CI = 1.13–2.67, adjusting for age and smoking).
Controlling for dietary variables that we have previously reported to be associated with bladder cancer risk in this cohort, ie, intakes of total fluid, 17 cruciferous vegetables, 18 and vitamin E, 19 did not alter the regional relative risks (data not shown). Although from Table 1 we observed that water intake was lower in the Northeast than in the West (2.2 vs. 2.8 cups/day), total fluid intake was similar in these two regions (1820 vs. 1867 ml/day) and thus, not surprisingly, controlling for total fluid intake had no effect on the regional relative risks.
Overall, residence in U.S. urban areas did not increase risk of bladder cancer compared with rural areas (residence determined at study baseline;Table 4). In contrast, within the states of New York and New Jersey, and in New England, individuals living in urban areas had an elevated risk of bladder cancer compared with those living in less populated areas (Table 5). In the state of New York, 77% of the cases were either living in New York City or on the heavily populated western end of Long Island, but only 46% of the NY State cohort members lived in those two regions. Controlling for urbanization did not affect the relative risks of bladder cancer by region. In a stratified analysis, controlling for age and smoking history, the relative risk for living in the Northeast, compared with the West, within strictly urban areas was 1.99 (95% CI = 1.23–3.20). The relative risk for the Northeast versus the West in areas that are not strictly urban (urban/rural) was 1.36 (95% CI = 0.81–2.29). We could not examine the “rural” stratum as there were insufficient numbers.
Regional Variation at Birth, Ages 15 and 25 Years
Participants in this study reported their state of residency at birth, ages 15 and 25 years (in the 1992 questionnaire). Elevated rates of bladder cancer were apparent in those who had lived in the Northeast at ages 15 and 25 years, compared with those who had lived in the West at ages 15 and 25 years (multivariate (MV) RR = 1.51, 95% CI = 1.11–2.04 at age 15; MV RR = 1.64, 95% CI = 1.21–2.23 at age 25). The multivariate relative risk of bladder cancer for men born in the Northeast was 1.34 (95% CI = 1.00–1.81) compared with those born in the West. Of those living in the Northeast at the age of 25, 21% still resided in the Northeast when we initiated this cohort study (age range of the men 40–75 years). Those men who were living in the Northeast at age 25 and at baseline had a relative risk of 1.17 (95% CI = 0.71–1.92) compared with those who had moved out of the Northeast at some point after 25 years of age. Men who lived in the Northeast at baseline, but not at 25 years old, had a relative risk of 0.66 (95% CI = 0.39–1.11) compared with men who lived in the Northeast at baseline and at age 25 years. Among those who lived in the Northeast at ages 15, 25 and in 1986, the relative risk of bladder cancer was 1.77 (95% CI = 1.15–2.71) compared with those who lived in the West at ages 15, 25 and in 1986, after controlling for history of smoking and current smoking status.
Greater incidence of bladder cancer may result if screening and detection efforts are more prevalent and tumors are detected at an earlier stage in the progression of the cancer. As medical procedures are slightly more frequently applied in the Northeast regions than in other regions of the United States (Table 1), early diagnosis of bladder cancer may be more common in those areas. Incidence of nonadvanced bladder cancer (stages 0 and I) in the Northeast was about 40% higher than that of the West (Table 6). In addition, incidence of advanced bladder cancer (stages II-IV) was substantially elevated in the Northeast when compared with other regions. The West had the lowest incidence rates.
In this cohort of male health professionals, a 65% increase in bladder cancer incidence was observed in the Northeast of the United States compared with the West, after adjusting for age and smoking history. In comparison, a 13% and 30% elevation in risk was observed in the South and the Midwest, respectively, compared with the West. The incidence of bladder cancer was highest in urban regions of New York State and New Jersey.
Excess mortality observed in urban areas in the northeastern United States, especially in northern New Jersey, has been hypothesized to be related to the high concentrations of chemical industries. 2 Because mortality rates are also relatively high in rural areas of the Northeast, studies have attempted to explain the excess mortality in the northern part of New York state and in New England, where chemical industrial exposures are less prevalent. 3,20 Elevated risks of bladder cancer were reported among textile and leather workers in rural New England 3 and among truck drivers in Vermont and New Hampshire. 20 The presence of regional differences for bladder cancer incidence in this cohort suggests that occupational exposures from “high risk” industries (eg, dye, leather, and rubber) cannot alone explain geographic variations in bladder cancer mortality in the United States, as it is unlikely that health professionals had industrial exposures.
In this cohort study, controlling for pack-years of smoking explained only some of the geographic variation, and is unlikely to account for the remaining excess incidence rates observed in the Northeast. Residual confounding by smoking could still account for the excess risk in the Northeast if there was substantial measurement error or if other aspects of smoking are important (eg, type of tobacco, filtered vs. nonfiltered cigarettes). Nevertheless, because never smokers had a 44% higher incidence of bladder cancer in the Northeast, compared with the West, cigarette smoking is unlikely to account for all of the excess risk observed in the Northeast.
Relative risks by region reported in this study were based on mailing addresses of participants at the time of recruitment (1986); however, these addresses may not necessarily represent lifetime exposures. Of the men living in the Northeast in 1986, 67% reported living in that region at ages 15 and 25, and 70% reported living in that region at age 25; restricting analyses to these stationary groups resulted in stronger associations for the Northeast. Our findings support the idea that duration of stay is an important determinant of bladder cancer risk in the Northeast. Similarly, our data do not strongly support the hypothesis that in vitro or childhood exposures are critical in and of themselves, as we would have expected the relative risks in the Northeast to be greater at birth and age 15 years than in the other analyses.
We did not observe an overall effect of urbanization on bladder cancer risk among the health professionals, although the excess risks in the states of New York and New Jersey were primarily observed in urban areas. When we examined incidence rates for each region by level of urbanization, we observed that within densely populated areas bladder cancer rates were almost twice as high in the Northeast (vs. West), but the excess risk was much smaller (36%) within the less densely populated areas. If air pollution is a risk factor for bladder cancer we would expect to see higher rates of bladder cancer in urbanized area throughout the United States, but we did not. We cannot rule out, however, the possibility that chemicals emitted from industries in urban parts of the Northeast may be initiating or promoting bladder carcinogenesis. Alternatively, some of the higher urban rates may be explained by residual smoking effects (cigarette smoking was more common in urban areas during the 1950s 9), or interactions between smoking and other environmental factors.
Previous studies have suggested that drinking water, perhaps serving as a vehicle for arsenic, chlorination byproducts, or other putative bladder carcinogens, may explain some of the excess risk of bladder cancer. 21 Although we do not have data on water supply contamination, we have previously shown that high water intake is associated with a decrease in the overall bladder cancer risk, and the trend was observed in each region individually. 17 Presently, there is considerable speculation (and on-going investigations) on whether water contaminants can explain some of the excess mortality of bladder cancer in New England. 22 Unfortunately, we cannot thoroughly address this issue in this study.
Other factors such as diet, religion, health profession, race/ancestry, marital status, and medical history did not explain the differences in bladder cancer risk across regions of the United States. Similarly, because the incidence rates of both early and advanced cases of bladder cancer are greater in the Northeast, and the risk was not elevated among those men who recently moved to the Northeast, it is unlikely that the excess rates observed resulted from better access to medical care and detection bias.
A few ecologic and observational studies have observed higher rates of bladder cancer in regions with either low soil selenium levels, 23,24 or individuals with low blood or toenail levels. 25–27 Selenium levels in the Northeast are substantially lower than in the Midwest (in crops 28 and in toenails 29), suggesting that perhaps differences in selenium intake, particularly 30–40 years ago when food transportation was less widespread, might explain some of the regional differences observed in the United States today.
Given the excess risk observed in this study of health professionals, it is clear that at least a portion of the excess risk cannot be attributed to latency effects of occupational exposures. While smoking accounted for a portion of the risk observed in the Northeast region, some excess risk remains that cannot be explained by differences in smoking patterns. Furthermore, dietary factors, lifestyle and medical practices could not account for the geographic variations observed in the health professionals.
1. Ries LAG, Miller BA, Hankey BF, Kosary CL, Harris A, Edwards BK. (eds). SEER Cancer Statistics Review, 1973–1997; Tables and Graphs. Bethesda, MD: National Cancer Institute, 1997.
2. Blot W, Fraumeni J. Geographic patterns of bladder cancer in the United States. J Natl Cancer Inst 1978; 61: 1017–1023.
3. Brown L, Zahm S, Hoover R, Fraumeni J. High bladder cancer mortality in rural New England (United States): an etiologic study. Cancer Causes Control 1995; 6: 361–368.
4. Mason T, McKay F, Hoover R, Blot W, Fraumeni J Jr. Atlas of cancer mortality for U.S. counties: 1950–1969. Vol DHEW Publ No: (NIH)76–780. Washington DC: US Government Printing Office; 1975.
5. Pickle L, Mason T, Howard N, Hoover R, Fraumeni J Jr. Atlas of U.S. cancer mortality among whites: 1950–1980. Vol DHEW Publ No. (NIH) 87–2900. Washington DC: US Government Printing Office; 1987.
6. Devesa S, Grauman D, Blot W, Pennello G, Hoover R, Fraumeni J Jr. Atlas of Cancer Mortality in the United States: 1950–94. Vol DHEW Publ No. (NIH) 99–4564. Washington (DC): US Government Printing Office; 1999.
7. McCredie M. Bladder and kidney cancers. Cancer Surv 1994; 19: 343–368.
8. Fraumeni JF Jr. Cigarette smoking and cancers of the urinary tract: geographic variation in the United States. J Natl Cancer Inst 1968; 41: 1205–1211.
9. Devesa S, Grauman D, Blot W, Fraumeni JF Jr. Cancer surveillance series: changing geographic patterns of lung cancer mortality in the United States, 1950 through 1994. J Natl Cancer Inst 1999; 91: 1040–1050.
10. Silverman D, Hartge P, Morrison A, Devesa S. Epidemiology of bladder cancer. Hematol Oncol Clin North Am 1992; 6: 1–30.
11. Hoover R, Cole P. Temporal aspects of occupational bladder carcinogenesis. N Engl J Med 1973; 288: 1040–1043.
12. Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett WC. Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol 1992; 135: 1114–1126.
13. U.S. Department of Agriculture. Composition of foods - raw, processed, and prepared, 1963–1992. Agricultural Handbook No. 8 Series. Washington, DC: Department of Agriculture, Government Printing Office, 1993.
14. Feskanich D, Rimm EB, Giovannucci EL, Colditz GA, Stampfer MJ, Litin LB, Willett WC. Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. J Am Diet Assoc 1993; 93: 790–796.
15. Dorkin TJ, Robson CN, Neal DE. The molecular pathology of urological malignacies. J Pathol 1997; 183: 380–387.
16. D’Agostino RB, Lee MLT, Belanger AJ, Cupples LA, Anderson K, Kannel WB. Relation of pooled logistic regression to time dependent Cox regression analysis: The Framingham Heart Study. Stat Med 1990; 9: 1501–1515.
17. Michaud D, Spiegelman D, Clinton S, Rimm EB, Curhan GC, Willett WC, Giovannucci EL. Intake of total fluid and specific beverages and risk of bladder cancer in U.S. men. N Engl J Med 1999; 340: 1390–1397.
18. Michaud D, Spiegelman D, Clinton S, Rimm E, Willett W, Giovannucci E. Fruit and vegetable intake and incidence of bladder cancer in a male prospective cohort. J Natl Cancer Inst 1999; 91: 605–613.
19. Michaud DS, Spiegelman D, Clinton SK, Rimm EB, Willett WC, Giovannucci E. Prospective study
of dietary supplements, macronutrients, micronutrients, and risk of bladder cancer in US men. Am J Epidemiol 2000; 152: 1145–1153.
20. Hoar S, Hoover R. Truck driving and bladder cancer mortality in rural New England. J Natl Cancer Inst 1985 771–774.
21. Cantor KP. Drinking water and cancer. Cancer Causes Control 1997; 8: 292–308.
22. Reynolds T. Prevention study turns spotlight on bladder cancer. J Natl Cancer Inst 1999; 91: 1102–1103.
23. Clark LC, Cantor KP, Allaway WH. Selenium in forage crops and cancer mortality in U.S. counties. Arch Environ Health 1991; 46: 37–42.
24. Shamberger RJ, Tytko SA, Willis CE. Antioxidants and Cancer Part VI. Selenium and age-adjusted human cancer mortality. Arch Environ Health 1976; 31: 231–235.
25. Schrauzer GN, White DA, Schneider CJ. Cancer mortality correlation studies - III: Statistical associations with dietary selenium intakes. Bioinorg Chem 1977; 7: 23–31.
26. Helzlsouer KJ, Comstock GW, Morris JS. Selenium, lycopene, a-tocopherol, b-carotene, retinol, and subsequent bladder cancer. Cancer Res 1989; 49: 6144–6148.
27. Nomura A, Heilbrun LK, Morris JS, Stemmermann N. Serum selenium and the risk of cancer, by specific sites: case-control analysis of prospective data. J Natl Cancer Inst 1987; 79: 103–108.
28. Kubota J, Allaway WH, Carter DL, Cary EE, Lazar VA. Selenium in crops in the United States in relation to selenium-responsive diseases of animals. J Agric Food Chem 1967; 15: 448–453.
29. Hunter D. Biochemical indicators of dietary intake. In: Willett WC, ed. Nutritional Epidemiology. 2nd ed. New York: Oxford University Press, 1998.